Greg Stark <gsstark(at)mit(dot)edu> writes:
> But I think this result is deceptive. It's still only 17% of the time. There
> doesn't seem to be a way to get the linux gprof to display the total time per
> function including children.
Er ... did you look all the way through the output? The standard gprof
output shows the flat timings you quoted followed by a call-tree report.
regards, tom lane